JPH02126212A - Focus detecting device - Google Patents

Abstract

PURPOSE:To shorten an arithmetic time while maintaining focus detection accuracy by calculating a correlation quantity at different shift quantity pitch nearby a focusing point and where a defocusing quantity is large. CONSTITUTION:A couple of subject image signals are inputted to a focus detection arithmetic means 30 through a focus detection optical system 24 and a photoelectric converting means 25 to calculate the defocusing quantity from a correlative shift value. Then, when the correlation quantity is found, its variation is set to 1st fine pitch within a 1st shift quantity range indicating that an image is almost in focus. Further, the pitch quantity is set larger than the 1st pitch except within the 1st shift quantity range where the absolute value of the defocusing quantity is large. Consequently, the pitch is made fine nearby the focusing point where accuracy is required, but the pitch is made large at a distance from the focusing point where the accuracy is not required so much to shorten the arithmetic time.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to a focus detection device for a camera or the like.

[Conventional side 1] A pair of subject images formed by light beams passing through different parts of a photographing optical system are photoelectrically converted by a pair of photoelectric conversion means to obtain a subject image signal consisting of a pair of discrete data, A focus detection device that performs a predetermined correlation calculation while relatively shifting the subject image signal of It has been known.

Such a focus detection device will be explained using FIGS. 6 to 13.

Figure 6 shows a conventional focus detection device with interchangeable lenses.
This shows an embodiment applied to an eye reflex camera, in which an exchangeable lens 10 can be detachably mounted on a camera body 20.

When the lens 10 is attached, the photographing light beam coming from the subject passes through the photographing lens 11 and reaches the camera body 20.
A part of the light is reflected upward by the main mirror 21 provided at the main mirror 21 and guided to a finder (not shown).

At the same time, another part of the photographing light flux passes through the main mirror 21 and is further reflected by the sub-mirror 22, thereby being guided to an autofocus module 23 (hereinafter referred to as AF module) as a focus detection light flux.

An example of the configuration of the AF module 23 is shown in FIG.

In Fig. 2, the AF module is the field lens 27.
and a C having a focus detection optical system 24 consisting of a pair of re-imaging lenses 28A and 28B and a pair of light receiving sections 29A and 29B.
It is composed of an OD (discharge coupled device) 25.

In the configuration as shown below, the ray of the photographing lens 11 +fj
A pair of regions 1 symmetrical with respect to the optical axis 17 included in the pupil 16
The light beams passing through 8A and 18B are formed as a primary image near the field lens 27, and further formed into a pair of secondary images on a pair of light receiving portions of the CCD 25 by the field lens 27 and the re-imaging lens 28A528B. When the secondary image coincides with a predetermined focal plane (not shown) which is a conjugate plane of the film, it is in focus, and the relative positions of the pair of secondary images in the light receiving section arrangement direction on the CCD 25 are not inspected 111. This is a predetermined value determined by the configuration of the optical system.

The pair of light-receiving sections 29A and 29B each consist of n light-receiving elements Ai and Bi (i=1 to n), and when the primary image coincides with the film conjugate plane, the corresponding light-receiving element (A+
and Bl, A, and B2...) are arranged so that the outputs thereof are equal.

When the imaging plane of the secondary image is located on a plane shifted from the film conjugate plane, the relative position of the pair of secondary images on the CCD 25 is in the direction of shift in the optical axis direction of the primary image (i.e. For example, in the case of the rear bin, the positional relationship between the pair of secondary images is relatively widened, and in the case of the front bin, Narrows down.

The light-receiving elements Ai and Bi forming the light-receiving sections 29A and 29B are composed of charge storage type elements such as fort diodes, and the light-receiving element output is determined by storing charge for a charge storage time corresponding to the illuminance on the CCD 25. The output level can be set to be suitable for processing.

Returning to FIG. 6 again, the explanation will be continued.

The control means 26 is a focus detection calculation means (AFCPU).
)30 baud) Receives tvM, R'all start and end commands from P+ and sends control signals according to the commands to CCD2
5 controls the start and end of charge accumulation in the CCD 25, and also gives a transfer lock signal to the CCD 25 to transfer the light receiving element output signal to the AFCPU in time series. Also, when the light receiving element output signal is generated, the sensor control means 2
6 sends a synchronization signal to P4 of the AFCPU 30,
The AFCPU 30 synchronizes with the generated synchronization signal and outputs the built-in AFCPU.
AD conversion was started by the D conversion means, and thereafter, at every cycle time of the transfer row 7, the photodetector output was sampled and AD converted by boat P3 to obtain AD conversion data (2n pieces) corresponding to the number of photodetectors. Thereafter, a known focus detection calculation (correlation calculation) to be described later is performed based on the data to obtain a defocus amount corresponding to the difference between the imaging plane of the primary image and the film conjugate plane.

The AFCPU 30 controls the display form of the autofocus display means (AF display means) 40 using P5 based on the focus detection calculation result.

For example, the triangular display section 41 is activated for the front bin, the triangular display section 43 is activated for the rear bin, the round display section 2 is activated for in-focus, and the rose display section 44 is activated for focus detection imbalance. Thus, the AFCPU 30 controls the AF display means 40.

The AF CPU 30 also controls the drive direction and drive amount of the AF motor 50 based on the focus detection calculation result, and moves the photographing lens 11 to the in-focus point.

The AFCPU 30 determines the sign of the defocus amount (front focus,
A drive signal for rotating the AF motor 50 in a direction in which the photographing lens 11 approaches the in-focus point is generated from P2 according to the rear lens. The rotational motion of the AF motor is transmitted through a body transmission system 51 comprised of gears and the like built into the body 20 to a coupling 53 on the body side provided at the mount portion of the body 20 and the lens 10.

The rotational motion transmitted to the coupling 53 on the body side is further transmitted to the lens transmission system 12, which is composed of a coupling 14 on the lens side mated thereto and gears built into the lens 10, and finally to the photographing lens. 11 moves in the focusing direction.

Further, the drive amount of the AF motor 50 is fed back from the boat Pl to the AFCPU 30 by converting the amount of rotation of gears, etc. of the body transmission system 51 into a pulse number by an encoder 52 constituted by a photointerrupter or the like.

That is, the AFCPU 30 controls the driving amount of the AF motor 50, that is, the number of pulses fed back from the encoder 52, in accordance with parameters such as the reduction ratio of the body transmission system 51 and the lens transmission system 12, so that the photographing lens 1
1 can be moved by a predetermined movement amount.

The AFCPU 30 has a built-in pulse counter for counting the number of pulses input from the boat PI, and a comparison register for comparing the contents of the pulse counter.
It has a function of generating an internal interrupt when the contents of the pulse counter and the comparison register match.

The AFCPU 30 controls the drive amount of the AF motor 50 in the following order. First, before starting driving the AF motor 50, the contents of the pulse counter are cleared and a desired number of pulses is set in the comparison register.

Next, driving of the AF motor 50 is started.

The rotation of the AF motor 50 causes the encoder 52 to generate pulses, which are counted up by a pulse counter.

When the contents of the pulse counter match the comparison register, an interrupt is generated and the AFCPU stops the AF motor by interrupt processing. In this way, the AF motor is driven and controlled by the desired number of pulses.

The lens 10 has a built-in lens CPU 13, and
Information necessary for the FCPU 30 (for example, AF-related information such as the number of rotations of the coupling 14 per unit movement of the photographic lens 11) is communicated on the body side via the lens side contact 15 and the body side contact 63 provided with the mount section. Send to bus 64.

The AFCPU30 receives the AFf! from the lens CPU13. The I-linked information is received from the boat P6 connected to the communication bus 64.

An overview of focus detection calculations such as correlation calculations in the AFCPU 30 will be described below with reference to FIGS. 8, 9, and 10.

The AFCPU 30 converts the COD data corresponding to the light-receiving elements Ai and Bi (i=1 to n) (in some cases, the difference) to at and bi (i=!-n).
), the correlation amount C is calculated by the correlation calculation shown in equation (1).
(L) is required.

C(L)=:Σ+ a(i+L)-b(i) l
==”(1)! However, in formula 1), L is an integer and is the relative shift amount (shift amount) in units of bifts of the light receiving element of a pair of CCD data, and the phase l511y4 calculation of formula (1) The range of the parameter i is determined as appropriate depending on Schift (ML) and f - the number of evenings n.

When the CCDCC data, bi, corresponds to a matrix, the combination of CCD data in equation (1), ie, the range taken by the parameter I, can be determined, for example, as shown in FIG. 10(A).

il 0 In figure (A), shift itl is -4 to +4
The combination of CCD data that is changed within the range of , and the calculation is performed at each shift amount by ``r pieces'' and ``0'' is represented by the digit t of the matrix L.

When set in this way, the range of parameter i is as shown in the following equation.

INT((L-1)/?3)≦i≦IN? ((L◆1
)/2-1n-2)---(2) It is also possible to determine the range of the parameter i as shown in FIG. 10(B).

In this case, the range of parameter i is as shown in the following equation.

L≦i≦n (L≧0), 15t≦n+L (L
<0)・”(3) The calculation result of equation (1) is a pair of Correlation 4C(L) becomes minimum in shift acid where the correlation of CCD data is high.

However, in reality -L, the relative shift 1L is due to the light receiving part 29A.
, 29B, the correlation amount C(L) itself also becomes discrete. Therefore, the correlation ic (L) was determined by calculation.
Not necessarily directly from Phase I! ! The minimum value of the quantity C(L) is not necessarily obtained.

Therefore, using the F method of three-point interpolation shown in Figure 4, the correlation itc
Find the minimum value C(x) of (L).

In other words, if the minimum value of the discretely calculated correlation amount C(L) is obtained when the relative shift iL is L=x, then the correlations corresponding to the relative shift amounts x-1, x, x+1 before and after that are obtained when the relative shift iL is L=x. fLc (L) is C(x −L), C(
x), C'(x+1), so first of all, the minimum correlation W:C(x) among the obtained data and the remaining two before and after
Draw a straight line H connecting the larger correlation amount (C(x+1) in Figure 9) of the correlation amounts C(x-1) and C(x+1), and then draw the next remaining correlation JiC(x-1) A straight line H passing through
A straight line J having an opposite slope is drawn to find the intersection W of these two straight lines H and J.

The coordinates of this intersection W are the correlation shift value X, and its correlation #C
(X.), and from these coordinates, it is possible to determine the correlation shift value X= that maximizes the correlation state and the minimum correlation amount C(Xs) for continuous shift amounts.

If this three-point interpolation method is expressed as an arithmetic expression, the correlation shift value X
, can be expressed as the following equation, and the correlation amount (x a is the following equation % equation % (5) Here, in equations (4) and (5), D can be expressed as the following equation.

Furthermore, in equations (4) and (5), 5LOP is the correlation Ic (x
-1), C(x), and C(x+1), which is the larger deviation, and can be expressed as follows.

Correlation shift adjusted by equation (4)) dominant x m is a pair of C
Representing the relative deviation amount of CD data, and assuming that the pitch of the light receiving element is y, the relative lateral deviation ig of the two subject images formed on the CCD 25 is g = yXx-... (
8) It can be expressed as follows.

Also, the defocus amount DEF at the focal plane is calculated by the following formula: %formula%() It can be approximated as follows.

Here, KX is a coefficient determined by the structural conditions of the focus detection optical system shown in FIG.

Furthermore, the larger the value of the parameter 5LOF obtained using equation (7), the deeper the indentation of the correlation amount C(L) shown in FIG. 8, that is, the larger the correlation.
This shows that DEF has high reliability.

Next, the Jlj-like shift measurement method will be explained.

The il1 diagram shows the combinations of CCD data at and bi used for correlation calculation as "book" and "O" on the matrix when the number of COD data n is set to 16 and the range of shift scale is set to -12 to +12. It is expressed. in this case,
a used in the correlation calculation of equation (1) for each shift 14L
There are four pairs of i and bi. When the extreme value of the correlation amount C(L) appears at iL=O, the photographing lens is in focus, and the defocus amount is approximately 0. As the shift amount moves away from O, the absolute value of the amount of defocus of the photographing lens increases.

Figure 12 shows shift stitching and correlations i, C(
This is a graphical representation of the relationship L). In this diagram,
Correlation 1c (L) shows the lowest value at shift ff1L=-1. It can also be seen that near the shift amount L=-1, the correlation amount C(L) drops sharply. The K shown on the prisoner's side represents the order in which correlation calculations are performed in Schiff)itL, and indicates that the calculation is performed by increasing L by 1 from L=-12 until =+12.

On the other hand, FIG. 13 is a graph showing the relationship between the shift amount and the correlation tc (L) when the subject image is formed with a large defocus for the same subject.

In this figure, when the shift amount is L=+6, the correlation amount C
(L) shows the lowest value of 1, and in the vicinity of shift 1L=6, as shown in FIG. L=-1
Even at 2°+12, the correlation tc (L) does not have a large value as shown in Fig. 12. In other words, when the subject image is formed with a large defocus, high spatial frequency components are removed and low frequency components are This is because since the image consists of only the components, the peak representing the correlation state becomes gentle.

Based on the defocus amount obtained in this way, the imaging position of the subject image through the photographing lens IT is sequentially monitored,
The photographing lens 10 was focused on the intended subject.

[Problems to be Solved by the Invention] In the prior art as described above, when a predetermined correlation calculation is performed while relatively shifting a pair of subject image signals, and when searching for the shift amount with the highest degree of correlation, If there are many combinations of relative shifts in the subject image signal, correlation calculations will be performed for all shift amounts, resulting in a huge amount of calculation and, as a result, an increase in calculation time. There was a problem in that the detection responsiveness decreased.

Moreover, the photographing lens is out of focus, and the absolute value of the defocus amount is large (i.e., shift) ML.
When searching for the extreme value of the correlation amount C(L) where the absolute value of is large, it is not necessary to take the variation of shift L finely. Also, if the absolute value of the defocus amount is large, there is no need to calculate the defocus amount with high precision.
In some cases, it is only necessary to detect whether the defocus is positive or negative, and for that reason, there is no need to set the pitch of the shift amount very closely near the focus.

On the other hand, in order to find the correlation shift value X that gives a highly accurate defocus amount in accordance with equation (4) near the focus, it was necessary to find the correlation amount C(L) using as fine a variation as possible in the shift amount. .

[Means for Solving the Problems] (a) The focus detection device of the present invention forms two beams of light that have entered and passed through the photographic lens from the subject on different optical paths, individually on the detection surface as a pair of subject images. Focus detection optical system.

(b) a pair of photoelectric conversion means each comprising a plurality of light receiving elements disposed on the detection surface and generating a subject image signal consisting of a discrete data group corresponding to a light intensity distribution of the subject image;

(C) Correlation shift that maximizes the correlation state between the pair of subject image signals by relatively shifting the pair of subject image signals from the pair of photoelectric conversion means by a predetermined shift amount and performing a correlation calculation. a focus detection calculation means for determining a value from these correlation calculations, and calculating a defocus amount corresponding to a difference between the position of the imaging plane of the photographing lens and the position of a planned focal plane conjugate to the film from the correlation shift value; Further, the focus detection calculation means sets the predetermined shift a at a first pitch when performing the correlation calculation in a range of a first shift ψ representing the vicinity of focus. However, when performing the correlation calculation outside the range of the first shift a, the predetermined shift amount is set at a pitch larger than the first pitch.

In addition, when performing the correlation calculation outside the range of the first shift amount, the focus detection calculation means sets the predetermined shift amount to a pitch larger than the first pitch, and The pitch of the shift i may be gradually increased as the amount moves away from the range of the first shift amount.

[Operation 1] In the present invention, since the focus detection device is configured as described above, in the range of the first shift amount that provides defocusing acid near the focus that requires precision, the change in shift amount ( However, outside the range of the 1st degree shift amount, which gives a large defocus amount away from the in-focus point where very precision is not required, the change in shift e (pitch) can be increased, and therefore the calculation It can save time.

[Example J FIG. 1 schematically shows a focus detection device fi100 of the present invention. Optical system etc. 1 that guides the image of the subject to the focus detection optical system 24
1.21.20 is similar to the prior art described above.

Two light beams from the subject reflected by a submirror 22 in the camera body are guided to a focus detection optical system 24 similar to that shown in the prior art, and each light beam forms a primary image and a secondary image.

The photoelectric conversion means 25' generates a pair of subject image signals corresponding to the light intensity distribution of the pair of secondary images. This is accomplished with a CCD 25 and sensor control means 26 similar to those shown in the prior art.

The focus detection @ calculation means 30' is the AFCPU shown in the prior art.
30, which corresponds to AD of the subject image signal.
Difference of subject image signal after AD conversion (differential coefficient of subject image signal with respect to shift)
Calculate. Furthermore, according to a predetermined program, a correlation calculation is performed by relatively shifting the pair of subject image signals by a predetermined shift amount, and a correlation shift value X that maximizes the correlation state between the pair of subject image signals is determined. It is determined from these correlation calculations, and from the correlation shift value X, a defocus amount corresponding to the difference between the position of the image forming plane of the photographic lens 11 and the position of the expected focal plane is calculated.

In the present invention, the focus detection calculation means (AFCPU
) 30' operation.

The operation of the AFCPU 30' in the first embodiment will be explained below with reference to FIG.

FIG. 2(A) is a part of the flowchart of the correlation calculation performed inside the AFCPU 30', and FIG. 2(B) shows a second This is obtained when the embodiment shown in Figure (A) is applied. This is a graphical representation of the relationship between the shift amount and the correlation amount C(L).

FIG. 2(A) is a flowchart that is applied when the camera is directed toward a subject and the photographic lens is at a certain focal position, and is used to sequentially monitor the imaging position based on external focusing commands. It is a flowchart.

Correlation calculation starts at a certain focus position, #100
Set the initial value of the shift amount to -12.

Next, in #105, the correlation amount C(L) is calculated for the set shift scale using (1) a chopstick.

In #110, it is tested whether the currently set shift amount has reached 12, and if it has reached 12, the correlation calculation loop is exited. If shift iL is not 12, shift in #115, test whether JIL satisfies either condition -4 or L≧4 (outside the range of first shift e), and if not, i.e. If the shift amount is close to in-focus, the increment Δ is set to 1 in #120, and the process proceeds to #130.

On the other hand, if the condition is satisfied in #115, that is, if the shift amount corresponds to a state where the absolute value of the defocus amount is large, the increment Δ is set to 2 in #125, and the shift amount is shifted to #130.
In #130, the shift scale is updated by adding the increment Δ to the current shift scale, and the process returns to #105 to repeat the above operations.

Finally, the correlation amount C(L
), the correlation calculation loop is exited at #11O. After that, select the one with the minimum (large) value from the obtained correlation amounts C(L), obtain the correlation shift value X, as explained in FIGS. 8 and 9, and finally calculate the defocus amount. Then, when the correlation calculation is performed in order from =1 to =17 as shown in Figure 2 (B), which is used to perform defocus display and lens drive as described above, the shift amount is also -12 accordingly. ,-10゜-8-6,-
4, -3, -2-02, 3° 4.6, 8, 10.12, the shift ML showing the lowest value of the correlation amount C(L) is If it is found within the range of variation l, the correlation shift value X- can be determined as described above. If the shift scale that indicates the lowest value of the correlation tc (L) is not found in the range of variation 1 (first pitch) around O, the shift scale corresponding to the focus is not found in the range of variation 1 (first pitch) around O. The correlation calculation may be performed again using the variation l (first pitch) to obtain the correlation 1tc (L), and the correlation shift value X- may be obtained in the same manner as above, or if the absolute value of the defocus amount is originally Since it is a large state, accuracy is not so necessary, and the shift scale that shows the lowest value can be directly used as the correlation shift value X.

A second embodiment of the present invention will be described below with reference to FIG.
′, etc. ) FIG. 3(A) is a part of the flowchart of the correlation calculation performed inside the AFCPU 30', and FIG. 3(B) shows a third calculation for the same state as when the correlation graph of FIG. 13 was obtained.
This is a graph representing the relationship between the shift iL and the correlation amount C(L) obtained when the flowchart shown in FIG.

In FIG. 3(A), first, in #200, the initial value of shift μL is set to 0, and the initial values of variables LP5 and LM are set to L, that is, O.

Next, in step #205, a correlation amount C(L) is calculated for the set shift error using equation (1) or the like.

In #210, it is tested whether the currently set shift amount has become -12, and if it has become -12, the CPU exits from the correlation calculation loop.

If it is not -12, use #215 to set the shift scale to L.
Test whether the condition of
Test whether P satisfies the condition of LP≧4, and if it satisfies the condition, that is, the LP in a state where the absolute value of the defocus field is large
', set the increment Δ to 2 in #225 and #2
Proceed to step 35.

If the variable LP does not satisfy the condition of LP≧4 in #220, that is, if the LP is near the focus, #23
0, the increment Δ is set to 1 and the process proceeds to #235.

In #235, the increment Δ is added to the current variable LP to set the variable L.
P is updated, the shift scale is roughly set to the variable LP in #236, the process returns to #205, and the above operation is repeated.

On the other hand, if the condition L>0 is satisfied in #215, that is, if L is in the previous bin state, the variable LM
Test whether LM≦-4 is satisfied, and if it is satisfied, that is, if the absolute value of defocus amount is large LM, set the increment Δ to -2 in #245 and perform step #255.
Proceed to.

If the variable LM does not satisfy the condition of LM≦-4 in #240, that is, if the LM is near the focus, #2
At 50, set the increment Δ to -1 and proceed to #255.

In #255, the increment Δ is added to the current variable LM to set the variable L.
M is updated, the shift scale is roughly set in the variable LM in #256, the process returns to #205, and the above operation is repeated.

Finally, the correlation amount C up to titL=-12
After calculating (L), the process exits from the correlation calculation loop in #210. After that, select the one with the minimum (large) value from the obtained correlation amounts C(L), and
As explained in the figure, the correlation shift value are performed sequentially from =1 to 17, and the shift amount is also changed accordingly from =1 to =17. -12, -2.
3. -3.4. -4.6. -6.8° fist...12. -1
It changes to 2.

If the shift scale indicating the lowest value of the correlation amount C(L) is found in the range of variation 1 (range of shift amount of the first pitch) around 0 (shift) corresponding to the focus, the above-mentioned The shift value X- can be immediately obtained as follows. Also, Figure 3 (B)
If the shift scale that shows the lowest value of the correlation amount C(L) is not found in the range of variation 1 (the range of the shift amount of the first pitch), the variation l( The correlation calculation is performed again with pitch l) and the correlation ψc (L
) and obtain the correlation shift value xI in the same manner as above, or alternatively, since the absolute value of the defocused acid is originally large, accuracy is not necessary, and the shift amount Lt that shows the lowest value - direct correlation shift value It can also be used as an Xs.

A third embodiment of the present invention will be described below with reference to FIG.
It is composed of etc. ) FIG. 4(A) is a part of the flowchart of the correlation calculation performed inside the AFCPU 30', and FIG. Seen when applying the flowchart shown in Figure (A),
This is a graph representing the relationship between the shift amount and the correlation IIC(L).

In FIG. 4(A), first, at #300, the initial value of 51L is set to -12, and the initial value of increment Δ is set to 5.

Next, in #305, the correlation amount C(L) is calculated for the set shift amount using equation (1) or the like.

In #31O, it is tested whether the currently set shift amount has reached 12, and if it has reached 12, exits from the correlation calculation loop. If it is not 12, test in #315 whether the shift amount satisfies either the condition -4 or L≧4, and if it is not satisfied, that is, if L is near the focus, # 325 with increment Δ as 1 #34
Proceed to o.

On the other hand, if the condition is satisfied in #315, that is, if the absolute value of the defocus amount is large, then #
At 320, test whether the shift amount "L is L<-4, and
If <-4, in step #330, 1 is subtracted from the current increment Δ to update the increment Δ. If the result in #320 is not −4, the process proceeds to #335, where 1 is added to the current increment Δ to update the increment Δ.

At #340, the shift scale is updated by adding the increment Δ to the current shift scale, and the process returns to #305 to repeat the above operation.

Finally, the correlation amount C(
L) is calculated and the correlation calculation loop is exited at #31O. Then, from among the correlation amounts C(L) obtained, the one with the minimum (large) value is selected, and as explained in FIGS. 8 and 9, the correlation shift value As shown in FIG. 4(B), correlation calculations are performed in order from =1 to 13, and the shift amount is also calculated accordingly. 12,-8,-5゜-3,-
2, -1, 0, 1, 2, 3, 5, 8 degrees 12. In this way, the first shift corresponding to the vicinity of the focus)
In the range of lL, the variation is 1, and outside the range of the first shift amount, as the absolute value of the shift amount increases, (
As one moves away from the first shift acid range), the variation was gradually increased to 2°3.4. The subsequent processing is performed in the same manner as in the previous embodiment.

A fourth embodiment of the present invention will be described below with reference to FIG.
It is composed of etc. ) FIG. 5(A) is a part of the flowchart of the correlation calculation performed inside the AFCPU 30', and FIG.
This is a graph representing the relationship between the shift amount and the correlation ic (L) obtained when the flowchart shown in FIG. 3(A) is applied.

In pJIJS diagram (A), first shift amount L at #400.
The correlation amount C(L
) is calculated. The order in that case is shown in WIJs diagram (B).
1 to 9, and the shift table-1L is
L=0.1. -1.2. -2. ．． 3°-3, 4-4, and the obtained correlation amount C(L) is shown as one line in FIG. 5 (CB).

In #405, it is tested whether a reliable minimum value of the correlation amount C(L) is found in the range of 13≦L≦3. That is, in order to obtain the above-mentioned correlation shift value x1, -4≦L≦4
It is necessary that a dent with the correlation IIC (L) shown in Figures 8 and 9 exists within the range of .

If a reliable minimum value of the correlation amount C(L) is found in #405 (actually, #400 and #405 operate in combination, it is not necessary to change the shift scale to -4, If the minimum value is found at L=0, then L=0.1.
-1 is sufficient to complete the correlation calculation), skip #410 and exit from the correlation calculation loop.

If a reliable minimum value of correlation TItC(L) is not found in #405, for example, in the case shown in FIG. 12 (B), proceed to #410 and set ML to -12≦L≦
-6, 6≦L≦12, the correlation amount C(L
) is calculated. The order in that case is shown in Figure 5 (B) = 10
17, and the shift scale is L=-12.
, -10°-8, -6, 6, 8, 10, 12,
The determined correlation uc (L) is indicated by # in FIG. 5(B).

The processing after #410 is the same as in the above-described embodiment, and will therefore be omitted.

In the gSl to fourth embodiments described above, the correlation result C(L
), in the range of shift) ML near focus, the variation of the shift amount is calculated as 1, and in the range of shift amount with a large absolute value of the defocus amount, the variation of shift) ML is calculated as 2 or more. However, the range of the shift amount near the focus (for example, in the embodiment of FIG. 2(A),
≦L≦4) may be changed manually or automatically.

For example, you can store the range of shift around the focus area for each interchangeable lens as lens information, and set the range of shift around the camera focus based on this information. The camera may set the range of the shift amount near the in-focus area based on lens information such as the focal length of the camera.

Also, various camera operation modes related to response time and calculation time (manual, one-shot or continuous, wide focus detection area, multi or spot, fast or slow motor winding speed, tracking operation for moving subjects) or not), change the range of the shift amount near the focus area, or set whether to change the shift amount variation between the focus area and other ranges. Good too.

[Effects of the Invention] In the present invention, when calculating the correlation amount C(L), the range of the first shift a near one focus and the range of the first shift amount where the absolute value of the defocus amount is large, Correlation calculations are performed by changing the amount of variation in the shift amount, so even if the focus detection area expands and the number of data increases, focus detection accuracy near focus can be maintained, and the absolute value of the defocus amount can be maintained. It becomes possible to detect a large shift amount without increasing the calculation time.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a focus detection device according to the present invention, and FIGS. 2(A) and 2(B) are a flowchart and an explanatory diagram of a first embodiment of the present invention. 3(A) and (B) are a flowchart and an explanatory diagram of a second embodiment of the present invention, and FIGS. 4(A) and (B) are a flowchart and an explanatory diagram of a third embodiment of the present invention. ff15 Figures (A) and (B) are a flowchart and an explanatory diagram of the fourth embodiment of the present invention, Figure 6 is a configuration diagram of a conventional focus detection device, etc., and Figure 7 is a diagram of a conventional technology. FIG. 8, fjS9, FIG. 10, FIG. 11, FIG. 12,
FIG. 13 is an explanatory diagram of focus detection calculation. [Explanation of symbols of main parts] 0...... Photography lens, 3...
・・・・・・・・・・・・・Lens CPU, 0・・・・
・・・・・・・・・・・・Camera body, 4・・・・・・
......Focus detection optical system, 5...
・・・・・・・・・CCD sensor 5′・・・・・・・・・
...Photoelectric conversion means, 0.30'...AFCPU (focus detection operator-stage) 0... AF display means, 0.
・・・・・・・・・・・・AF motor. 00・・・・・・・・・Focus detection device. Figure L Flute 2 da CB) Figure 3 (A) Figure 3 CB) Scale Figure 4 (B) To L Figure 5 (B) Injection rise Figure 7 a2 Negative S gθ bζ (A) Whistle /2 figure

Claims (1)

[Scope of Claims] 1. A focus detection optical system that forms two beams of light that have entered and passed through a photographing lens from a subject on different optical paths as a pair of subject images on a detection surface, each of which is formed on a detection surface. a pair of photoelectric conversion means that is composed of a plurality of arranged light receiving elements and that generates a subject image signal that is composed of a discrete data group corresponding to the light intensity distribution of the subject image; and a pair of subjects from the pair of photoelectric conversion means. A correlation calculation is performed by relatively shifting the image signals by a predetermined shift amount, a correlation shift value that maximizes the correlation state between the pair of subject image signals is determined from these correlation calculations, and the correlation shift value and a focus detection calculation means for calculating a defocus amount corresponding to the difference between the position of the imaging plane of the photographic lens and the position of a planned focal plane conjugate to the film, wherein the focus detection calculation means comprises: , when performing the correlation calculation within a range of a first shift amount representing the vicinity of focus, the predetermined shift amount is set at a first pitch, and the correlation calculation is performed outside the range of the first shift amount. In the case where the focus detection device performs the focus detection, the predetermined shift amount is set at a pitch larger than the first pitch. 2. When performing the correlation calculation outside the range of the first shift amount, the focus detection calculation means sets the predetermined shift amount to a pitch larger than the first pitch, and The focus detection device according to claim 1, wherein the pitch of the shift amount is gradually increased as the amount moves away from the range of the first shift amount.